xref: /openbsd-src/gnu/usr.bin/perl/dist/Unicode-Normalize/t/test.t (revision 4e1ee0786f11cc571bd0be17d38e46f635c719fc)
1
2BEGIN {
3    unless ('A' eq pack('U', 0x41)) {
4	print "1..0 # Unicode::Normalize cannot pack a Unicode code point\n";
5	exit 0;
6    }
7    unless (0x41 == unpack('U', 'A')) {
8	print "1..0 # Unicode::Normalize cannot get a Unicode code point\n";
9	exit 0;
10    }
11}
12
13BEGIN {
14    if ($ENV{PERL_CORE}) {
15        chdir('t') if -d 't';
16        @INC = $^O eq 'MacOS' ? qw(::lib) : qw(../lib);
17    }
18}
19
20#########################
21
22use strict;
23use warnings;
24BEGIN { $| = 1; print "1..72\n"; }
25my $count = 0;
26sub ok ($;$) {
27    my $p = my $r = shift;
28    if (@_) {
29	my $x = shift;
30	$p = !defined $x ? !defined $r : !defined $r ? 0 : $r eq $x;
31    }
32    print $p ? "ok" : "not ok", ' ', ++$count, "\n";
33}
34
35use Unicode::Normalize;
36
37ok(1);
38
39sub _pack_U   { Unicode::Normalize::pack_U(@_) }
40sub _unpack_U { Unicode::Normalize::unpack_U(@_) }
41
42#########################
43
44ok(NFD(""), "");
45ok(NFC(""), "");
46ok(NFKD(""), "");
47ok(NFKC(""), "");
48
49ok(NFD("A"), "A");
50ok(NFC("A"), "A");
51ok(NFKD("A"), "A");
52ok(NFKC("A"), "A");
53
54# 9
55
56# don't modify the source
57my $sNFD = "\x{FA19}";
58ok(NFD($sNFD), "\x{795E}");
59ok($sNFD, "\x{FA19}");
60
61my $sNFC = "\x{FA1B}";
62ok(NFC($sNFC), "\x{798F}");
63ok($sNFC, "\x{FA1B}");
64
65my $sNFKD = "\x{FA1E}";
66ok(NFKD($sNFKD), "\x{7FBD}");
67ok($sNFKD, "\x{FA1E}");
68
69my $sNFKC = "\x{FA26}";
70ok(NFKC($sNFKC), "\x{90FD}");
71ok($sNFKC, "\x{FA26}");
72
73# 17
74
75sub hexNFC {
76  join " ", map sprintf("%04X", $_),
77  _unpack_U NFC _pack_U map hex, split ' ', shift;
78}
79sub hexNFD {
80  join " ", map sprintf("%04X", $_),
81  _unpack_U NFD _pack_U map hex, split ' ', shift;
82}
83
84ok(hexNFD("1E14 AC01"), "0045 0304 0300 1100 1161 11A8");
85ok(hexNFD("AC00 AE00"), "1100 1161 1100 1173 11AF");
86
87ok(hexNFC("0061 0315 0300 05AE 05C4 0062"), "00E0 05AE 05C4 0315 0062");
88ok(hexNFC("00E0 05AE 05C4 0315 0062"),      "00E0 05AE 05C4 0315 0062");
89ok(hexNFC("0061 05AE 0300 05C4 0315 0062"), "00E0 05AE 05C4 0315 0062");
90ok(hexNFC("0045 0304 0300 AC00 11A8"), "1E14 AC01");
91ok(hexNFC("1100 1161 1100 1173 11AF"), "AC00 AE00");
92ok(hexNFC("1100 0300 1161 1173 11AF"), "1100 0300 1161 1173 11AF");
93
94ok(hexNFD("0061 0315 0300 05AE 05C4 0062"), "0061 05AE 0300 05C4 0315 0062");
95ok(hexNFD("00E0 05AE 05C4 0315 0062"),      "0061 05AE 0300 05C4 0315 0062");
96ok(hexNFD("0061 05AE 0300 05C4 0315 0062"), "0061 05AE 0300 05C4 0315 0062");
97ok(hexNFC("0061 05C4 0315 0300 05AE 0062"), "0061 05AE 05C4 0300 0315 0062");
98ok(hexNFC("0061 05AE 05C4 0300 0315 0062"), "0061 05AE 05C4 0300 0315 0062");
99ok(hexNFD("0061 05C4 0315 0300 05AE 0062"), "0061 05AE 05C4 0300 0315 0062");
100ok(hexNFD("0061 05AE 05C4 0300 0315 0062"), "0061 05AE 05C4 0300 0315 0062");
101ok(hexNFC("0000 0041 0000 0000"), "0000 0041 0000 0000");
102ok(hexNFD("0000 0041 0000 0000"), "0000 0041 0000 0000");
103
104ok(hexNFC("AC00 11A7"), "AC00 11A7");
105ok(hexNFC("AC00 11A8"), "AC01");
106ok(hexNFC("AC00 11A9"), "AC02");
107ok(hexNFC("AC00 11C2"), "AC1B");
108ok(hexNFC("AC00 11C3"), "AC00 11C3");
109
110# 39
111
112# Test Cases from Public Review Issue #29: Normalization Issue
113# cf. http://www.unicode.org/review/pr-29.html
114ok(hexNFC("0B47 0300 0B3E"), "0B47 0300 0B3E");
115ok(hexNFC("1100 0300 1161"), "1100 0300 1161");
116ok(hexNFC("0B47 0B3E 0300"), "0B4B 0300");
117ok(hexNFC("1100 1161 0300"), "AC00 0300");
118ok(hexNFC("0B47 0300 0B3E 0327"), "0B47 0300 0B3E 0327");
119ok(hexNFC("1100 0300 1161 0327"), "1100 0300 1161 0327");
120
121ok(hexNFC("0300 0041"), "0300 0041");
122ok(hexNFC("0300 0301 0041"), "0300 0301 0041");
123ok(hexNFC("0301 0300 0041"), "0301 0300 0041");
124ok(hexNFC("0000 0300 0000 0301"), "0000 0300 0000 0301");
125ok(hexNFC("0000 0301 0000 0300"), "0000 0301 0000 0300");
126
127ok(hexNFC("0327 0061 0300"), "0327 00E0");
128ok(hexNFC("0301 0061 0300"), "0301 00E0");
129ok(hexNFC("0315 0061 0300"), "0315 00E0");
130ok(hexNFC("0000 0327 0061 0300"), "0000 0327 00E0");
131ok(hexNFC("0000 0301 0061 0300"), "0000 0301 00E0");
132ok(hexNFC("0000 0315 0061 0300"), "0000 0315 00E0");
133
134# 56
135
136# NFC() and NFKC() should be unary.
137my $str11 = _pack_U(0x41, 0x0302, 0x0301, 0x62);
138my $str12 = _pack_U(0x1EA4, 0x62);
139ok(NFC $str11 eq $str12);
140ok(NFKC $str11 eq $str12);
141
142# NFD() and NFKD() should be unary.
143my $str21 = _pack_U(0xE0, 0xAC00);
144my $str22 = _pack_U(0x61, 0x0300, 0x1100, 0x1161);
145ok(NFD $str21 eq $str22);
146ok(NFKD $str21 eq $str22);
147
148# 60
149
150## Bug #53197: NFKC("\x{2000}") produces...
151
152ok(NFKC("\x{2002}") eq ' ');
153ok(NFKD("\x{2002}") eq ' ');
154ok(NFKC("\x{2000}") eq ' ');
155ok(NFKD("\x{2000}") eq ' ');
156
157ok(NFKC("\x{210C}") eq 'H');
158ok(NFKD("\x{210C}") eq 'H');
159ok(NFKC("\x{210D}") eq 'H');
160ok(NFKD("\x{210D}") eq 'H');
161
162ok(NFC("\x{F907}") eq "\x{9F9C}");
163ok(NFD("\x{F907}") eq "\x{9F9C}");
164ok(NFKC("\x{F907}") eq "\x{9F9C}");
165ok(NFKD("\x{F907}") eq "\x{9F9C}");
166
167# 72
168
169